Beruflich Dokumente
Kultur Dokumente
HEATSINKANALYTICAL
MODELING
Masterthesis
JoaquimGuitartCorominas
March2011
Tutor:AmirARZAND
Dpartementd'ElectrotechniqueetdeSystmesd'Energie
coleSuprieuredlectricit
INTRODUCTION.....................................................................................................................................3
HEATGENERATIONINELECTRONICDEVICES..........................................................................................4
2.1
2.2
IGBT.......................................................................................................................................................4
DIODE....................................................................................................................................................5
INTRODUCTIONTOHEATTRANSFER......................................................................................................7
THERESISTANCESINAHEATSINKMODEL.............................................................................................7
4.1
SINKAMBIENTRESISTANCERSA......................................................................................................................8
4.2
RBFPLATECONDUCTIONRESISTANCE...............................................................................................................9
4.2.1 Conductionheattransfer................................................................................................................10
4.3
RSPRESISTANCE........................................................................................................................................11
4.4
RFARESISTANCE........................................................................................................................................11
4.4.1 Finconductionfactor:
.................................................................................................12
4.4.1.1
4.4.1.2
4.4.1.3
4.4.2
4.4.3
Convectioncoefficienthc.................................................................................................................16
Radiationequivalentcoefficienthr..................................................................................................16
4.4.3.1
Efficiencyanalysis................................................................................................................................13
Rectangularlongitudinalfinsdemonstration.........................................................................................13
Trapezoidallongitudinalfinsequations..................................................................................................15
Radiationinheatsinks............................................................................................................................16
INTRODUCTIONTOFLUIDSDYNAMICS................................................................................................19
5.1
DEVELOPINGANDFULLYDEVELOPEDFLOW....................................................................................................21
5.2
CONVECTIONCOEFFICIENTINNATURALCONVECTION.......................................................................................22
5.2.1 Parallelplates..................................................................................................................................22
5.2.2 NaturalconvectioninUchannel.....................................................................................................22
5.2.2.1
5.2.2.2
5.3
WorkofYovanovich................................................................................................................................23
WorkofBilitzky.......................................................................................................................................24
CONVECTIONCOEFFICIENTINFORCEDCONVECTION.........................................................................................25
SIMPLEALGORITHM............................................................................................................................27
MULTIPLEHEATSOURCEMODEL.........................................................................................................29
MULTIPLEHEATSOURCESALGORITHM................................................................................................32
COMPUTATIONRESULTS.....................................................................................................................34
9.1
DISCUSSIONOFTHERESULTS.......................................................................................................................34
10 OPTIMIZATIONPROCESS.....................................................................................................................37
10.1
10.2
THEGENETICALGORITHM...........................................................................................................................37
OPTIMIZATIONSAMPLEFORHEATSINKS........................................................................................................39
11 CONCLUSION.......................................................................................................................................41
12 REFERENCES........................................................................................................................................42
APPENDIX1.PROFILESANDCOMPUTATIONRESULTS.................................................................................43
APPENDIX2.AIRPROPIETIESANDEQUATIONS............................................................................................48
APPENDIX3.MULTIPLEHEATSOURCEALGORITHMFORMATLAB................................................................50
1 Introduction
Electronicshasleadedmosttechnologicaladvancesofthepast60years.
There are technologies with domains particularly developed for
electronics such as material science, electromagnetism, system
dynamicsandalsoheattransfer.
The relation to heat transfer is because the heat generation of
electronics devices. Commonly, these devices need additional
cooling in order to avoid extreme temperatures inside it. Heat
sinks allow this supplementary cooling, so they are omnipresent
inelectronicassemblies.
Heatsinkcanworkbyforcedconvection,naturalconvectionorliquidcooling.Normallyinelectronic
assembliestheyaremadeofmaterialswithgoodthermalconductionsuchasaluminumorcopper.
Theheattransferinsinksisespeciallybyconvection,butalsobyradiation.Radiationheattransfer
canrepresentupto30%ofheatrateinnaturalconvectionheatsinks.
The manufacturing process is usually by extrusion,
butalsobycast,bonded,folded,skivedandstamped
processes.
Therearealotofgeometriesavailableandtheyare
generally adapted to each specific requirement.
However, a very common heat sink profile is the
rectangular parallel fin one. This profile forms U
channels,wheretheconvectionphenomenonisable
tobemodeledbyempiricalcorrelations.
Theradiationprocessisalmostageometricproblem,soitsanalysiswillbeaminororderstudy.
Themodelingofrectangularparallelfinheatsinksallowsananalyticalstudy.Thisstudycanleadto
determiningtheparametersofaheatsinkforaspecificapplication,mainlyforelectronicsindustry.
Theheattransferprocessesthatoccurinaheatsinkarestudiedinthiswork.Thereisalsoproposed
an algorithm for rectangular parallel fin heat sinks and the computation study of its results. These
computationresultsarecomparedtoafiniteelementprogramsolutioninordertoknowtheerrorof
theproposedmodel.Finally,issuggestedanoptimizationapplicationforthisalgorithm.
2 Heatgenerationinelectronicdevices
The power electronic devices are made with PN structures that have heat losses when current
circulates across it. The overheating can destroy the PN structure, in which junctions are the most
sensiblepart.
TherearesomephenomenonsthatproduceheatlossesinPNstructures.
Conduction losses. When a PN junction allows the current circulations, there is a potential
dropbetweenthePNterminalsthatproduceheat.Theheatgeneratedinconductionstatus
canbeexpressedas:
(1)
Blockinglosses.Aresidualcurrentremainswhenthedeviceisinblockingposition.
Switchinglosses.Thecurrentandvoltageswitchesarenotinstantaneous.WhenaPNdiode
is in a conduction state and is going to be a blocking state a transient negative current is
present as the blocking voltage is being applied: it is called recovery phenomenon.
Therefore, the switching frequency is an important factor to take into account when heat
lossesareanalyzed.
Drivinglosses.
2.1 IGBT
OnstateLosses
StaticLosses
BlockingLosses
TotalPowerLosses
DrivingLosses
TurnonLosses
SwitchingLosses
TurnoffLosses
Fig.1Powerlossesinelectronicsdevices.
Becausetheyareonlycontributingtoaminorshareofthetotalpowerdissipation,forwardblocking
lossesanddriverlossesmayusuallybeneglected.
Onstatepowerdissipations(Pfw/T)aredependenton:
Loadcurrent(overoutputcharacteristicvCEsat=f(iC,vGE))
JunctiontemperatureTj/T(K)
DutycyclesDT
4
Forgivendriverparameters,theturnonandturnoffpowerdissipations(Pon/T,Poff/T)aredependent
on:
Loadcurrentvd(V)
DClinkvoltageiLL(A)
JunctiontemperatureTj/T(K)
Switchingfrequencyfs(1/s)
Thetotallosses
foranIGBTcanbeexpressedas:
Powergeneratedper
switchon(W)
Powergeneratedper
switchoff(W)
(2)
Where:
,
Onstatepower
dissipation(W)
Where
, , / istheheatgeneratedperswitchon(J),
, , / istheheat
/
/
istheaverageloadcurrent(A),DTisthetransistordutycycle,
generatedperswitchingoff(J),
,
istheCollectoremittersaturationvoltage(V)
2.2 DIODE
Becausetheyareonlycontributingtoaminorshareofthetotalpowerdissipation,reverseblocking
power dissipations may usually be neglected. Schottky diodes might be an exception due to their
hightemperatureblockingcurrents.
Turnonpowerdissipationsarecausedbytheforwardrecoveryprocess.Asforfastdiodes,thisshare
ofthelossesmaymostlybeneglectedaswell.
Onstatepowerdissipations(Pfw/D)aredependenton:
Loadcurrent(overforwardcharacteristicvF=f(iF))
JunctiontemperatureTjD
DutycyclesDD
ForagivendriversetupIGBTcommutatingwithafreewheelingdiode,turnoffpowerlosses(Poff/D)
dependon:
Loadcurrent
5
DClinkvoltage
JunctiontemperatureTj/D(K)
Switchingfrequencyfs(1/s)
(3)
Where:
Powergeneratedper
switchoff(W)
Forwardpower
dissipation(W)
Where
istheheatgeneratedperswitchingoff(J),
istheaverageload
current(A),DDisthetransistordutycycle.
3 Introductiontoheattransfer
The heat sinks are elements that prevent the destruction of electronic equipment because of its
overheating.Themostcriticalpartinanelectronicdeviceisthesemiconductorjunction.Thejunction
temperaturecantexceedatemperaturegivenbythemanufacturer.
The heat sinks have different shapes depending on the nature of the coolant fluid (natural air
convection cooling, forced air convection cooling, liquid cooling...), the manufacturing process, the
electronicmodulepackaging
Tofacilitatetheunderstandingofheattransferlawsforelectricengineers,itcanbeusefultoexplain
thisasananalogybetweenelectricalandthermalresistances.
The Ohms law describes the relation between the current I, the potential difference V and the
resistanceRbetweentwopointsas:
(4)
In the thermal analogy, the potential difference V (V) is associated to temperature difference
between two points T (C), the electrical resistance R () is associated to a thermal resistance R
(C/W),andthecurrent(A)isassociatedtoaheatfluxratioQ(W).
(5)
The number of resistances of a model depends on the desired precision. A high precision model
requiresalargenumberofresistances.However,ahighnumberofresistancesinamodelcanreduce
significantlythecomputationspeed.Consequently,thereisacompromisebetweenthecomputation
speedandtheprecisionoftheresults.
Generally, the heat ratio Q is determined by the operating conditions of the semiconductor. The
temperatureincreaseT,isalsodeterminedbythemaximumjunctiontemperatureandthehighest
ambienttemperatureinhypotheticalextremeambientconditions.Therefore,usingtheequation(5),
thehighestthermalresistanceofanassemblyisfixed.
4 Theresistancesinaheatsinkmodel
The goal of the analysis is to determine the heat sink geometry and a device setup which allow
enoughheatdissipationforagivendevicesandworkingconditions.
The heat sinks can be meshed by many 3D thermal resistances which can involve a complex
modeling.Forsimpleanalyticalanalysis,thereisnomorethanoneheatsourceinvolved;itisuseful
tousetheonedimensionalmethodofequivalentresistances.
Inthislinealsystem,theglobalresistanceRcanbedividedintothreethermalresistancesRsa,Rcsand
Rjc(fig2).TheadditionofthesethreeresistancesistheglobalresistanceR,giveninequation(6).
(6)
Rjcistheresistancebetweenthesurfaceofthe
deviceandthejunctionofthesemiconductor.
Itisusuallygivenbythedevicemanufacturer.
Rcs
Ts
Tc
Rsa
Tj
Ta
Rjc
Packaging
Bond
(termal
grase)
plate
fins
Fig.2Resistancesinaheatsink
4.1 SinkambientresistanceRsa
TheanalysisleadstoadivisionoftheheatsinkresistanceRsaintothreesubresistances.
(7)
Rbf is the resistance due to the limited conduction of a flat plate when a uniform flux flows
perpendicularlytoitssurface.
Rspisaresistanceduetothefluxspreadingpenalization.Whenafluxflowsthroughaplatefroma
heat source area S1 to a dissipation area S2 and S2>S1, then the flux flow is not completely
perpendicularlyandspreadingresistanceappears.
Rfaistheresistancebetweentheplatesurfacethatsupportsthefins,andtheambient.Theheatis
drivenawayduetoconvectionandradiationheattransfer.Thefinshavealsoaconductiveresistance
asaresultofitsfinitethermalconductivity.
Alltheseresistanceswillbeexplainedaccuratelyinthenextsections.
tp
Convection
heattransfer
Conduction
throughthefins
Heatsource
Ambient
plate
fins
Radiationheat
transfer
Conduction
spreadingzone
Fig.3Heatsinkcommonfluxes
4.2 RbfPlateconductionresistance
Itisdirectlycalculatedusingtheequation5
(8)
Where k is the plate conductivity (W/Km), A is the heat transfer area (m2) and tp is the plate
thickness.
As an introduction to conduction heat transfer, the demonstration of equation (8) is developed
below.
4.2.1 Conductionheattransfer
The following equation (9) is deduced from the energy balance in a control volume in Cartesian
coordinates.
(9)
Where k is the plate conductivity (W/Km), in the internal heat generation (W/m3), in the
materialdensity(Kg/m3),Cpistheheatcapacityofthematerial(J/KgK),Tisthetemperatureandtis
thetime.
Insteadystateconditions,
0,inuniformfluxflowthroughxaxis
0and
inmaterialswithoutheatgeneration =0,theequation(9)isreducedtoequation(10):
0
(10)
Integrating the equation (10) and applying the boundary conditions T(0)=T0 and T(L)=TL the
temperaturedistributionthroughthewallcanbeexpressedas:
(11)
TheFourierLawforondimensionalfluxcanbeexpressedas:
(12)
Where qx is the heat flux (W) in x direction, k is the plate conductivity (W/Km) and A is the heat
transferarea(m2).
Thederivateofequation(11)iscalculatedas:
,
(13)
Substitutingtheexpression(13)inFouriersequation(12):
(14)
Finally,thethermalresistanceRbfisdeducedas:
10
,
,
(15)
4.3 Rspresistance
The Rsp resistance is due to the flux
spread through the plate thickness. A
platehastwosides.Inonesidethere
plate
is the heat source, and in the other
side there are the fins that dissipate
the heat. The source area is usually
smallerthanthedissipationarea.The
fluxdirectioninsidetheplatebecomes
notperpendicular to the surface and Fig. 4 Platespreading
itinvolvesaresistanceassociated.
Surface1
tp
Surface2
TheworksofYovanovichandAntonettileadtothefollowingexpressionforaheatsourcecenteredin
heatsinksurface:
1.410
0.344
0.043
0.034
(16)
Where is the ratio between the heat transfer surface 1 and the heat transfer surface 2, k is the
1.
plateconductivity(W/Km)andaisthesquarerootofsurface1:
4.4 Rfaresistance
The resistance between the plate surface that supports the fins and the environment is the Rfa
resistance.Thisresistanceincludesconduction,convectionandradiationheattransfer.TheNewtons
lawofcooling(17)isalinearexpressionthatcanbeusedtofindtheresistanceRfain(18).
(17)
Whereqistheheattransferrate(W),histheconvectioncoefficient(W/Km2),Aistheheattransfer
surface(m2),Tsisthesurfacetemperature(K)andTambistheambienttemperature(K).
1
(18)
However, this expression does not include the conduction resistance through the fins and the
radiationheattransfer.Thatwillleadtomodifytheexpression(17)intotheexpression(19)inorder
toincludeallheattransferphenomenon.
(19)
11
Where q is the heat transfer rate (W), hc is the convection coefficient (W/Km2), hr is the radiation
equivalentcoefficient(W/Km2),Apistheprimaryarea(m2),Afistheextendedarea(m2),isthefin
efficiency,Tsistheplatesurfacetemperature(K)andtheTambistheambienttemperature(K).
Allthesevariablesandcoefficientswillbeexplainedinthenextsections.
4.4.1 Finconductionfactor:
Fins have a finite conductivity. It means that the temperature can vary along its surface. But in
equation (19) the Ts is included as a fixed value, it is not a function like Ts(x). For this, the fin
efficiency is included to penalize the temperature variation along the fin surface without affecting
the linearity of equation (19). This efficiency will modify the affected surface Af of the fin, but the
primary plate surface Ap will retain the expected temperature Ts and it will be not affected by the
efficiency(fig5).
Finsurface
Plate
Primarysurface
Fig.5Surfacesinaparallelplateheatsink
Theefficiencydependsonthefingeometry.Inindustrytherearealotoffinprofilesavailable(fig7),
from pin fins to hyperbolic profile longitudinal fins. A rectangular profile fits accurately the most
longitudinalprofiles(fig6),soitwillbeusedintheproposedanalyticalmodel.
Fig.7Rectangularprofile,hyperbolicprofile,triangularprofile,
trapezoidalprofile
Fig.6Superpositionofa
trapezoidalandahyperbolic
profilefin.
12
4.4.1.1 Efficiencyanalysis
FromFouriersequations(12)andfromenergybalanceequations(9),isdeduced:
1
(20)
dAs
qx
dqconv
Ac(x)
qx+dx
dx
x
z
y
x
Fig.8Finprofile
4.4.1.2
Rectangularlongitudinalfinsdemonstration
Tam
Inthiscase,thesectionAcisconstant.
If P is assumed as the section
perimeter:
tab
Tb
(21)
(22)
Ac
L
Replacing
H
x
Asecondorderdifferentialequationis
obtained:
Fig.9Rectangularlongitudinalprofile
13
(23)
Therestrictionsareadiabatictip(x=H)andfixedbasetemperatureTb.
(24)
Thesolutionofequation(23)applyingrestrictions(24)is:
cosh
cosh mH
(25)
WhereHisthefinhigh(m).
Toknowthepowerdissipated,theheattransferatthefinbaseisanalyzed.
(26)
Substituting
forthederivativeofequation(25):
tanh mH
(27)
Theefficiencyistheratiobetweenthemaximumheatratethataperfectfincandissipateandthe
heat rate that dissipate a real fin. The maximum power that can dissipate a perfect fin is deduced
fromtheNewtonslawofcooling(17),expressedas:
(28)
WhereAfthefinareaandTbisthefinbasetemperature.
Fromequation(27)andequation(28)isdeducedthefinefficiencyassumingtab<<L.
tanh mH
tanh
mH
(29)
0substitutingtheparameterHin(29)for
Thisequationisalsovalidforaconvectivefintip,
Hc.
2
14
4.4.1.3 Trapezoidallongitudinalfinsequations
Asmentionedbefore,thetrapezoidalfinprofilewillbeusedinthemodelingofheatsinksduetoits
fitnesstomostfinprofiles.
Theequationsthatdeterminetheefficiencyofthefinsare:
(30)
Where:
/2
(31)
atan
(32)
sin
(33)
1 tan
2tan
(34)
1 tan
2tan
(35)
AllparametersusedaredefinedintheFig.10bellow:
tab
taf
H
Hc
Fig.10Trapezoidallongitudinalfinprofile
15
4.4.2 Convectioncoefficienthc
The convection is a phenomenon that allows the heat exchange between a solid and a fluid. Even
thoughtheNewtonslawofcooling(17)isalinearexpression,theanalysisofthecoefficienthand
the surface temperature Ts can become so complex. The reason is that the convection coefficient
dependsbasicallyonthetemperature,thegeometryandthefluxregime,butarealsurfacecannot
haveaconstanttemperature.
Thechapter5explainsbasicnotionsoffluiddynamics,whicharerequiredtostudytheconvection.
4.4.3 Radiationequivalentcoefficienthr
All bodies emit electromagnetic radiation due its temperature. The heat transferred by radiation
dependsonthebodytemperatureandtheambienttemperature.Theheattransferrateqrforabody
canbeexpressedas:
(36)
Where A is the body surface (m2), is the surface emissivity (dimensionless), is the Stefan
Boltzmannconstant(W/m2K4),Tsisthesurfacetemperature(K)andTambistheambienttemperature
(K).
So,itisdeducedbytheequation(36)thatthemostinfluentparameteristhetemperaturedifference
betweenthesurfaceandtheambient.
ThevalueoftheStefanBoltzmannconstantisgiveninSIby:
5.6704
10
(37)
Theemissivityisusuallygivenbythemanufacturer,eithergivingthefinishtype(degreased,black
anodize,clearanodize)eithergivingdirectlythesurfaceemissivityvalue.
Theequation(36)isproposedforsurfaceswhichonlyemitradiationtothesurrounding.However,
someshapesformsincludesurfaceswhichemitradiationtothesurroundingsandtoothersurfaces
simultaneously.Theequationofthesegeometriesincludesaviewfactor.
TheviewfactoristheproportionofallthatradiationwhichleavesasurfaceAandstrikessurfaceB.
4.4.3.1 Radiationinheatsinks
The heat sinks geometry is complicated concerning the radiation analysis. Some surfaces exchange
radiationeachotherandtotheambientsimultaneously.Othersurfacesexchangeradiationonlyto
the ambient. Therefore, the radiation in heat sinks involves more parameters than expressed in
equation(36).
ThearticleofYounesShabany[1],simplifiestheradiationinheatsinks.Theheattransferrateqr(W)
canbeexpressedas:
(38)
16
Wherenaisthenumberoffins,qch,ristheheattransferrateinchannelsurfaces(W),Adistheareaof
all surfaces those radiation does not strike other surfaces (m2), is the body emissivity, is the
StefanBoltzmann constant (W/m2K4), Ts is the surface temperature (K) and Tamb is the ambient
temperature(K).
Adcanbededucedby:
(39)
Allparametersofequation(39)aredefinedinFig.11.
taf
H
tp
L
tab
sb
Fig.11HeatSinkgeometricparameters
Theheattransferrateofachannelcanbewrittenas:
2
,
(40)
WhereFssurristheviewfactorfromthechannelsurfacetothesurroundings
Assuminganerror,Fssurrcanbewrittenas:
2
2
1
1
1
.
(41)
Where:
17
Andsmistheaverageinterfinspace.
Inordertoknowtheequivalentconvectioncoefficienthr,isproposedthenextequation(42):
(42)
Where Ap and Af are the surfaces (m2) defined in Fig. 5 and qr (W) is the heat transfer rate by
radiation.
18
5 Introductiontofluidsdynamics
Fluiddynamicsanalysesthebehaviorofafluidwhenforcesareappliedtoit.
u=0.99u
Nominallimitoftheboundary
layer
turbulent
Transitionregion
laminar
Fluidvelocityinsidetheboundarylayer
Fig.12Velocityregimesthroughasurface
Whenafluxflowsoverafixedsurface,thefluidparticlesspeedthenexttothesurfaceiszero.These
particlesalsoaffecttothecontiguousparticles,andaspeedgradientappears.Thenominallimitof
thezoneaffectedbythesurfaceisalayercalledvelocityboundarylayer,wheretheparticleshave
the99%ofthefreestreamflowvelocity.
The movement of the particles next to the surface edge is orderly. This is the laminar flow zone,
wherethevelocitiesvectorsareparalleleachother.Fromacertainpoint,theparticlesmovements
becomechaotic.Thisistheturbulentzone.
Between the laminar and turbulent zone, there is a transition zone. In this region, the first
turbulencesappear,andafterthat,thefluidiscompletelyturbulent.
Inthelaminarzone,theviscousforcesarestrongerthanthemovementforces.Contrary,afterthe
transitionszonethemovementforcesexceedtheviscousones.
TheReynoldsnumberReisaratiobetweenthemovementforcesandviscousforces.
(43)
Where isthemeanfluidvelocity,Lisacharacteristiclineardimension(m)andisthekinematic
viscosity(m2/s).TheReynoldsnumberdeterminesthefluidsregimeforagivensurface.
19
Similarly, the thermal behavior of the fluid has a boundary layer. The nominal limit of the thermal
boundarylayeristhe99%ofthefreestreamfluidtemperature.
Tfreestreamtemperature
Nominallimitoftheboundarylayer
T=0.99T
Ts surfacetemperature
Fig.13Fluxthermalbehavior
Thethermaldiffusivitydeterminestheadjustmentspeedofthefluidtemperaturetothatoneof
theirsurroundings.Itcanbecalculatedby:
(44)
Wherekisthefluidthermalconductivity(W/Km), isthefluiddensity(Kg/m3)andCpisthefluid
heatcapacity(J/Kg).
ThePrandtlnumberPristheratioofmomentumdiffusivitytothermaldiffusivity.Itcanbewritten
as:
(45)
Where isthedynamicviscosity(Kg/ms).Therelationbetweenkinematicanddynamicviscositycan
beexpressedas:
(46)
TheNusseltnumberNuistheratioofconvectivetoconductiveheattransfer.Itcanbeexpressedas:
(47)
20
WhereLisacharacteristiclineardimension(m).
TheGrashofnumberistheratioofflotationforcesandviscousforces.Thisnumberisverysignificant
innaturalconvectionproblems.Itcanbewrittenas:
(48)
Where g is the gravity acceleration (m/s2), is the thermal expansion coefficient (1/K), Ts is the
surfacetemperatureandLisacharacteristiclineardimension(m).
TheRayleighnumberistheproductofGrashofandNusseltnumbers.Itdeterminesthetypeofheat
transfer in a fluid. In low range values the heat transfer is basically by conduction. After a critical
value,theconvectionheattransfertypepredominates.
(49)
5.1 Developingandfullydevelopedflow
Theseareconceptsthatwillbetakingintoaccountinmostconvectionmodels.
Fully
developed
flow
Developing
flow
Developing
flow
plates
Boundarylayer
Fig.14Boundarylayersindevelopedandfullydevelopedflow
In developing flow between two parallel plates, the boundary layers do not touch each other.
Contrary, in the fully developed flow the layers are mixed and the nominal limit of the boundary
layerdisappears.Thesetworegimesareruledbydifferentequations,whichareusuallycombinedin
one expression. In low range Rayleigh numbers, the fully developed flow domains, while in large
rangeRayleighnumbers,thedevelopingflowdomains.Thedevelopingflowregionlengthdepends
onthefluidtype,theflowconditionandthegeometry.
21
5.2 Convectioncoefficientinnaturalconvection
Thenaturalconvectionoccurswhenasurfaceheatsfluidparticles.Theseparticlesbecomemoreor
less dense than the environment ones. Under the gravity action, the less dense fluid particles rise,
andthemoredenseparticlesfall.Therefore,arelativemovementbetweentheheatandcoldflowis
inducted.
5.2.1 Parallelplates
Elenbaas [2] modeled semi empirically the convection coefficient between parallel plates and
ambient.Theconvectioncoefficientcanbeexpressedas:
(50)
Wherekfisthefluidconductivity,
istheaverageNusseltnumberandsmistheinterfinspace.
Specificallyinthecaseofisothermalparallelplates,theNusseltnumberisdefinedas:
576
/
2.87
/
(51)
Where
length.
is the Rayleigh number with sm as a characteristic linear dimension and L is the fin
Then,theconvectioncoefficienthisdeductedbytheequation(40)withsmasacharacteristiclinear
dimension.
The equation (51) is a composite solution between the fully developed equation and the isolated
plateones,wherethedevelopingflowdomains.
5.2.2
NaturalconvectioninUchannel
taf
H
tp
L
w
Fig.15Heatsinkparameters
tab
sb
22
Intheheatsinks,thefinsarelocatedonaplate(Fig.15)andthereisalsotheinfluenceoftheplatein
theflow.Therefore,aconvectionmodelforconvectionalUchannelheatsinksisneeded.
5.2.2.1 WorkofYovanovich
Thearticle[3]ofYovanovich(1995),proposesamodelforrectangularheatsinks,inarelativelarge
Rayleighnumbervalues.Thecharacteristiclengthistakenasthesquarerootofthewettedsurface
( ).Thisistheareaoftheheatsinkwhichisintouchwiththeairflow(fig16).
Therefore, the convection coefficient
canbeobtainedfrom:
Wettedarea
(52)
Where
Fig.16Wettedarea(S)
TheNusseltnumberisdefinedby:
(53)
Where
isthebodygravityfunction,and
istheRayleighnumberwith asacharacteristiclength
(m).
Therefore,theRayleighisdefinedas:
(54)
AllproprietiesareevaluatedattheaveragefluidtemperatureTm.Inmostproblems,thetemperature
canbecalculatedas:
WhereTsisthefintemperatureand
Thediffusivelimit
(55)
istheambienttemperature(K)
isdefinedas:
23
3.192
1.868
1.189
(56)
WhereL,H,Warethedimensionsofthecuboidassociatedtotheheatsink(m)definedinFig.15.
TheuniversalPrandtlfunctionisdefinedas:
0.670
0.5
(57)
WherePristhePrandtlnumber.
Thebodygravityfunctioncanbewrittenas:
(58)
Where:
2
taistheaveragefinthickness(m),tafisthetipfinthicknessandtabisthebasefinthickness(m),and:
Wherenaisthenumberoffinsoftheheatsink.Theothergeometricparametersaredefinedinfig
15.
5.2.2.2 WorkofBilitzky
The studies of Bilitzky [4] (1986) improved the work [5] of Van de Pol and Tierney (1973). Both
studies are based on Elenbaas correlation for parallel plate arrays. The Bilitzky and Tierney
expressionsincludeaspectratiofactorsforabetterfittingofUchannelprofiles.Inthesecorrelations,
thecharacteristiclengthisthehydraulicradius.
(59)
WhereAisthechannelcrosssectionalarea(m2)andPisthechannelwettedperimeter(m2).
Theconvectioncoefficientisdefinedas:
(60)
Wherekfisthefluidconductivity(W/mK).
TheNusseltnumber,accordingtoVandePolworkcanbeexpressedas:
24
(61)
Where is the Van de Pol geometric parameter and Elr is the Elenbaas number. The Bilitzky B
geometricparametercanreplacetheVandePolones.
TheBilitzkyBparameteriscalculatedby:
24
(52)
1.25 1
1
0.483
9.14
0.61
WhereSmistheaverageinterfinspacingandHisthefinheight.
Inchannelproblems,theElenbaasnumberwithrasacharacteristiclengthisdefinedas:
(63)
Where
(m).
istheRayleighnumberwithrasacharacteristiclength(m)andListheheatsinklength
TheRayleighnumberiscalculatedby:
(64)
All fluid proprieties have to be calculated at the surface temperature Ts, with the exception of ,
whichistakenattheaveragefluidtemperatureTmusingtheequation(55)
Finally,theconvectioncoefficienthcanbecalculatedfromequation(60).
5.3 Convectioncoefficientinforcedconvection
The forced convection is a type of heat transfer in which fluid motion is generated by an external
source(likeapump,fan,suctiondevice,etc.).Theconvectioncoefficientsareremarkablyhigher,up
tox10times,sotheheatsinksdimensionscanbenotablyreduced.
25
The convection coefficient is taken from the model proposed by P. Teertstra in his article [6]
publishedin2000.Itcanbewrittenas:
(65)
WhereNuistheNusseltnumber,kfisthefluidconductivityandsminterfinspaceascharacteristic
lineardimension(m).
In parallel plate channels the flux is found developing conditions, fully developed conditions or
simultaneouslyinbothconditions.ChurchillandUsagi[7]proposedacompositesolution.
/
(66)
Where Nu is the composite Nusselt solution, Nufd is the asymptotic solution for a fully developed
flow,Nudevistheasymptoticsolutionforadevelopingflowandnisacombinationparameterwhich
dependsonthemodel.InthemodelproposedbyTeertstra,nhasthevalueof3.
All fluid temperatures will be calculated at film temperature Tf (K). The film temperature is the
averagetemperatureoftheboundarylayer.
WhereTsisthesurfacetemperatureand
(67)
(K)istheambienttemperature(K).
TheNusseltNumberforthefullydevelopedflowasymptotecanbeexpressedas:
1
2
(68)
TheNusseltnumberforthedevelopingflowasymptoteisdefinedas:
.
0.664
3.65
.
(69)
Wheresmistheaverageinterfinspace,PristhePrandtlnumberdefinedin(45),ResmistheReynolds
numberdefinedin(43)withsmasacharacteristiclineardimension(m)andListheheatsinklength
(m).
26
6 Simplealgorithm
The objective of this algorithm is to know the junction temperature Tjun for a given heat sink and
workingconditions.
Thegeneralexpressionusedtoknowthepowerdissipatedbyaheatsinkcanbewrittenas:
(70)
Where:
Tjun
Tamb
Rcs
Rjc
Rsa
Rbf
Rsp
Rfa
Thejunctiontemperaturecanbewrittenas:
(71)
All parameters can be deduced directly from the sink geometry and its proprieties, except the Rfa,
whichalsodependsonthesinksurfacetemperatureTs.
Therefore,determiningthesurfacetemperatureisarequirementtoknowthejunctiontemperature.
However,thesurfacetemperatureisimplicitinmodelfunctionsandisneededaniterationprocessin
ordertoknowthistemperature.
So,forRfa,thegeneralequationcanbewrittenas:
(72)
Assuming:
Allheattransferrateqoccursinfinsideplatesurface.
TheTstemperaturewillbetakenastheaveragesurfacetemperature.
Theconvectioncoefficientisconstantalongthefinssurface.
27
Theradiationequivalentcoefficientisconstantalongthefinssurface.
Atthispoint,theiterationprocesscanbedefinedas:
Ts1 80C
Ts 0C
q P thetotalheatpowerfromthedevice
While |Ts1Ts| 0.01
Ts Ts1
CalculatehcforTsfromanymodelproposed.
CalculatehrforTsfromthemodelproposed.
Calculatehtotas
Calculateforhtotfromthemodelproposed.
CalculateRfaas
CalculateTs1as
End
Ts Ts1
CalculateRfaby
Theotherresistancescanbedirectlycalculatedfromtheexplicitequationsdefinedinthe
previoussectionsandthedatagivenbythemanufacturer.
Usingequation 68 ,Tjunisdetermined.
Finally Tjun is compared with the max junction temperature Tjunmax value given by the
manufacturerinordertoknowthefeasibilityoftheheatsinkdesign.
28
7 Multipleheatsourcemodel
In heat sinks with multiple heat
sources, the lineal resistance method
becomes useless. The reason is that
are induced multiple asymmetric
resistancesinsidetheplate(Fig.16).
Finsidesurface
Device3
Tj3
Ra3
Rfa
Ta
Device2
Tj2
Device1
Tj1
Anunfeasiblecaseisexposedbelow.
Ra2
Ts
Ambient
Ra1
Fig.17multipleheatsourcesresistances
(73)
Where Tjun1 is the junction temperature of the device 1 (K), q1 is the heat transfer rate from the
device1(W),Ra1istheplateresistanceassociatetoheatflux1(C/W).
In some configurations, the average surface temperature Ts might be higher than a junction
temperature Tn. Then, for an eventual heat source device (qn>0), the resistance Ran, would be
negative. This is fiscally impossible, so the lineal resistance method is rejected as a method for
multipleheatsources.
ThesolutionisproposedinthearticleofY.S.Muzychka,J.R.Culham,M.M.Yovanovich[8].Their
modelsolvesmathematicallythemultipleeccentricheatsourcesquestion.
FromtheLaplaceequation:
T
x
T
y
T
z
(74)
Restricting:
/
ConductionatsourcesurfaceAs=Ldxwd
Platethicknessisolated
Sourcesideplatesurafceisolated
Convectiononthefinsideplatesurface
0
0
, ,
Wherehmismodifiedconvectioncoefficient,expressedas:
29
(75)
WhereApistheprimarysurface,Afisthefinsurface(Fig.5),isthefinefficiency,wisthesinkwidth
andListhesinklength(Fig.15).
Tofindthesolutionofequation(71)isusefulapplyingthisequality:
(76)
,
2
2
4
cos
sin
cos
sin
(77)
cos
sin
2 cos
sin
Where:
,
cos
cos
sin
2
2
sin
2
2
cos
cos
30
sin
sin
2
2
cos
cos
cos
cos
16
sinh
cosh
cosh
sinh
Wherekistheplateconductivity(W/mK), isthepowerofthedevicei,
geometricparametersfordedevicei(fig19).
Pd/A
ydi
Devicesidesurface
plate
Finsidesurface
,L,ware
Ldi
wdi
tp
hm
Tamb
Insulatingzone,withoutheattransmission.
Fig.18Heatsinkinsulatingzone
xdi
Fig. 19 Heatsinkdevicedistributionparameters
31
8 Multipleheatsourcesalgorithm
The objective of this algorithm is to know the junction temperatures for a given sink and working
conditions.Inthiscase,severalheatsourcescanbeaddedinanydistribution.
In order to know the junction temperatures of each
device, the multiple heat source model defined in the
previoussectionwillbeused.
Assumptions
102
101,5
101
100.5
Heatsinkdevicesurface
Ts1 80C
Fig. 20 Devicetemperaturedistributionsample
Ts 0C
q P1 P2 PN Totalheatpowergeneratedfromdevices
While |Ts1Ts| 0.01
Ts Ts1
CalculatehcforTsfromanymodelproposed.
CalculatehrforTsfromthemodelproposed.
Calculatehtotas
Calculateforhtotfromthemodelproposed.
CalculateRfaas
CalculateTs1as
End
Ts Ts1
32
Calculate
matrix
theambientfortheheatsourcejis:
, . canalsobeobtained
fromtheadditionofthecolumnj.
Calculate from
From the average temperature of the source j, the junction temperature is calculated
from:
,
,
,
ComparetheTjun,jforthegivenmaxjunctiontemperatureforthedeviceTjunmaxj.
33
9 Computationresults
Naturalconvectionhasbeentestedinordertoknowtheprecisionofthealgorithmsproposed.There
is available in the net a free online modeling tool, Rtools [9] , offered by Mersen. This program is
based on an advanced threedimensional numerical model, and there are many sink profiles
available.
The multiple heat source algorithm has included the models of Bilitzky and Yovanivich for natural
convection.Moreover,theBilitzkymodelhasbeenanalyzedwiththefluidproprietiesevaluatedat
the wall temperature and at the boundary temperature. The Van de Pol article, which the Bilitzky
modelis based,holds thewalltemperatureasthe referencetemperaturetocalculating mostfluid
proprieties.Anyway,itcanbeanexperimenttoverifyit.
Ithasbeenanalyzed45profilesof102.TheprofilesexcludedareeitherthosearenotavailableinR
toolsorarenotabletobeparameterizedbecauseitsgeometry.
Eachprofilehasbeentestedatlengthof50%,100%150%and200%ofitswidth.Then,ithasbeen
180tests,4perprofile.
Each simulation had got one heat source, which has occupied totally the sink surface. The source
heatgenerationratehasbeenestimatedforeachsimulationinordertohavejunctiontemperatures
from 90 C to 140 C. These values are usually the maximum junction temperature given by the
manufacturers.
Then,thealgorithmmodelandtheRtoolsresultshavebeencompared.Theresultsarepresentedin
theappendix1.
9.1 Discussionoftheresults
Theadmissibleerrorhasbeensetupat15%.Theappendix1showstherelativeerrorofeachmodel
respectthertoolsresults.
Theaverageerrorforthethreeproposedmodelsisexposedbelow:
BilitzkyWalltemperaturemodel
Bilitzkyboundarytemperaturemodel
Yovanovichmodel
Average
10,4%
10,5%
20,5%
Median
8,5%
8,7%
15,5%
Table1Errorcomparison
These results do not reveal an important average error difference between both Bilitzky models.
However,thereisaconsiderabledifferenceforsomeindividualtestsresults.Forcertainprofiles,the
differencebetweenbothBilitzkymodelsisashighas10%.
Eachprofilehasamodelwhichapproachbettertortoolresults.Of45profiles,theaverageerrorfor
the 4 test that each profile is subjected is minimal in BilitzkyWall model 20 times, in Bilitzky
Boundarymodel13timesandinYovanovichmodel12times.So,thereisnotanobviouswinner.
Ifthebettermodelforeachindividualprofilewasused,theaverageerrorwoulddropto8.12%and
themedianwouldfallto6.05%.
Inconclusion,amixedmodelcouldimprovetheprecision.
34
There is a remarkably correlation between the error from a given model and some geometric
parameters.
Someoftheseparametersare:
:istherelationbetweentheaverageinterfinspacing(sm)andthewettedarea(wa).
ThisparameteriswellcorrelatedtobothBilitzkymodelsasshowedinfig21andfig22.
: is the average interfin spacing. It is strongly correlated with the error of Yovanovich
modelasshowsfig23.
Fig.21.ErrorBilitzkywallvsLOG10(sm/wa).R2=0.528
Fig.22ErrorBilitzkyBoundaryvsLOG10(sm/wa).R2=0.625
35
Fig.23ErrorYovanovichvssm.R2=0.859
There is available in Matlab two applications, curve fitting tool and surface fitting tool, which may
helptoidentifynewcorrelations.
36
10 Optimizationprocess
An optimization process leads to find a set of parameters which minimize or maximize a function,
satisfyingasetofconstraints.
The optimization in heat sinks can minimize the volume, the price, the length or the weight for a
givenworkingconditions.
Anoptimizationisoftendevelopedasaniterationprocessruledbyanalgorithm.
The proposed convection model can be optimized, admitting a precision error. However, the
precision can be remarkably improved adding constraints to allow the algorithm work in favorable
rangeofparameters.
For example, if the error in Yovanovich model is lower at large interfin spacing values (fig 22). A
hypothetical constraint limiting the minimum interfin space could be added in order to work in
propitiousrangesofvalueswheretheerrorisstatisticallynotablylower.
Thereareseveralalgorithmsavailable.Inthenextsectionisthegeneticalgorithm.
10.1 Thegeneticalgorithm
Thegeneticalgorithmisbasedonbiologicevolution.Heatsinks Lowerlimit
have several parameters (Length, fin height, width) which
determine the sink performance. These parameters (P1, P2)
defineanindividualandareboundedbyupperandlowerlimits,
which are predetermined by physical, manufacture or model
limitations(fig24).
Upperlimit
Thereisalsoasetofconstraints.Theseconstraintsinvolvemore
than one parameter and must be evaluated after defining an
individual.
Theconstraintsmayinclude:
P1
P2
P3
P4
P5
P6
Fig. 24 Algorithmparameters
Geometric constraints: the devices must be placed in the sink surface, and cannot be
superimposed.
Manufacturing constraints: the heat sinks are usually made by extrusion. In extrusion, the
aspect ratio (the relationship between the spacing and the height of the fins) and the
minimumfinthicknessarelimited.Themanufacturingmethodshavebeennotablyimproved
during the 90s and 00s. Actually, most manufacturers offer aspect ratios up to 1:10 and
minimumfinthicknessof1mm[10].
Temperature constraints: The junction temperature cannot exceed the maximum junction
temperaturegivenbythemanufacturer.Inaddition,somestandardsinindustrycanrestrict
thetemperatureofthesinksurface.
Finally there is as objective function. This function specifies the variable (volume, length, weight)
whichiswantedtobeminimizedormaximized.
Thealgorithmstepsareexplainedbellow.
37
1.Creationofanindividualwhosatisfiesallconstraints.
2. Random creation of the first generation. All individuals are composed from the original
individual.
3.Orderingtheindividuals.Everyindividualhasanobjectivefunctionvalue.Moreover,they
mayormaynotsatisfytheconstraints.Theyareorderedfrombesttoworst.Thosethatmay
not satisfy the constraints obtain automatically a very high number as objective function
value.
()
Satisfytheconstraints.Ordered
frombesttoworst
Donotsatisfytheconstraints.
Extremevalue.
4.Crossoverandmutation.Thenextgenerationiscreatedeitherfromthecrossoverofthe
previous generation, either from the mutation of some parameters. Each algorithm has a
mutationratiofixedbytheuser.
()
Individualscreatedfromcrossover
Mutatedindividuals
5.Orderingtheindividualsas3.Iterationprocess.
6.Thealgorithmfinisheseitherafteracertainnumberofiterations,eitherafterobservingno
changesinthebestindividual(thefistafterordering)duringafixednumberofiterations.
38
10.2 Optimizationsampleforheatsinks
The multiple heat sources algorithm has been tested in Matlab. Each individual is composed by 3
continuousparameters(Fig.25):
L:Length
DPD: distance between pair of
devices
DLD:distancebetweenthelinesof
devices.
DPD
DPD
Devices
DPD
DLD
DPD
DPD
DPD
Figure25 HeatSinkOptimizationparameters
0,22
0,215
Length(mm)
0,21
0,205
0,2
0,195
0,19
0,185
0
50
100
Iteration
150
200
Figure26Lengthvariationalongiterations
39
0,014
0,012
DPD (mm)
0,01
0,008
0,006
0,004
0,002
0
0
50
100
150
200
Iteration
Figure27DPDVariationalongIterations
0,6
0,5
DLD (mm)
0,4
0,3
0,2
0,1
0
0
50
100
150
200
Iteration
Figure28DLDVariationalongIterations
40
11 Conclusion
Good agreement obtained in comparisons between the Rtool data and the algorithm proposed,
revealtheviabilityofusingthemultipleheatsourcealgorithmasatoolforheatsinks.
Threesampleapplicationsforthisalgorithmareexposedbellow:
Determiningthefeasibilityofaheatsinkdeviceassembly.
Optimizingheatsinkgeometryforaspecificapplication.
TracingResistanceLengthcurvesforagivenprofile.
The 180 tests of 45 profiles give significant information about the behaviour of the algorithm. The
analysisofthisbehaviourisakeyforimprovingfuturemodels.Inthisdocument,theerrorhasbeen
correlatedtotwoparameters,neverthelessanewstudycouldfindeasilynewcorrelations.Limiting
theactuationrangeoftheproposedalgorithmmayleadtoaprecisionincrease.
For a meticulous future work, the errors for natural convection, radiation and spreading
phenomenonmodelsmustbeanalyzedandparameterizedseparately.
Theconvectionmodelsmustbecomparedwithexperimentaldataandtheresultsofcomputational
fluid dynamics software such as Fluent or as Floworks. That could help to analyse thoroughly the
conductofeachmodelinfunctionofparticularparametersorrelations.
The proposed radiation model is an approximation in which the error depends on geometric
relations.Workingwithrealviewfactorsmayleadtoimprovetheradiationheatratioagreement.
Afterwards, mixing models or even creating new ones could make possible more performing
algorithms.
However,thesurfacetemperaturewillalwaysbeaninconvenientsinceitisnotaconstantvalue.It
changes along the plate surface and along the fin surfaces. Long heat sinks tend to have a great
varianceintheirsurfacetemperature.Theairproprietieschangeasitisheatedalongthesink.The
fluidheatabsorptioncapacitydecreasesandthehighestzonesoftheheatsinksarelesscooled.
Thiscreatesasymmetriesthatcannotbeeasilysimulatedbyanalyticalmethods.Asimple3Dmodel
whereonlytheplatewasmeshedcouldbeagoodwaytoremarkablyimprovetheprecision,without
wastingtoomuchcomputationalsources.
By using analytically approaches, other types of shapes could be also deduced. Some heat sink
profiles are very branched, but there are linear transformations available that can facilitate the
modelling.ItisalsonecessarytodevelopnaturalconvectionmodelsforVchannelstodeterminethe
convectioncoefficientforthesebranchedsinks.
41
12 References
[1]
Shabany,Younes(2008).Simplifiedcorrelationsforradiationheattransferrateinplatefin
heatsinks.http://www.electronicscooling.com/2008/08/
[2] Elenbaas,W(1942).Heatdissipationofparallelplatesbyfreeconvection.PhysicaVolIX,No
1.128.
[3] J. Culham, M Yovanovich and Seri Lee (1995). Thermal modeling of isothermal cuboids and
rectangular heat sinks cooled by natural convection. IEEE Transactions on components,
packaging,andmanufacturingtechnologypartA.Vol.19,NO3.September1995.
[4] Bilitzky, A. (1986). The Effect of Geometry on Heat Transfer by Free Convection from a Fin
Array, M.S. thesis, Department of Mechanical Engineering, BenGurion University of the
Negev,BeerSheva,Israel.
[5] Van de Pol, D. W. and Tierney, J. K. (1973). Free Convection Nusselt Number for Vertical U
ShapedChannels,J.HeatTransfer,87,439.
[6] P. Teertstra, M.M. Yovanovich and J.R. Culham. Analytical forced convection modeling of
PlatefinheatsinksJournalofelectronicsmanufacturing,Vol.10,No.4(2000)pp.253261.
[7] Churchill,S.W.andUsagi,R.,Ageneralexpressionforthecorrelationofratesoftransferand
otherphenomenon.A.I.Ch.E.Journal,Vol.18,pp.11211128.1972
[8] Y. S. Muzychka, J. R. Culham, M. M. Yovanovich (2008). Thermal Spreading Resistance of
Eccentric Heat Sources on Rectangular Flux Channels. Journal of Electronic Packaging JUNE
2003,Vol.125.
[9] http://www.rtools.com/
[10] Kuzmin,G.(2004)Innovativeheatsinkenableshotuplinks.Powerelectronicstechnology,pp
1420.November2004.
42
Appendix1.Profilesandcomputationresults
Bil.boundary
Bil.wall
79,6
99,7
21,2%
76,8
100,9
22,0%
75,6
102,6
22,1%
97,5
26,5%
111,5
19,6%
121,0
17,0%
120,2
128,4
16,3%
15,2
129
Width(mm)
30,48
30,5
124
Height(mm)
16,51
45,7
127
61
129
Code
61080
19,8
124
97,4
Width(mm)
39,61
39,6
10
135
108,6
Height(mm)
19,05
59,4
15
145
115,3
79,2
20
154
fins
Code
Yovanovich
23,3%
66171
Bil.Wall
(C)
Tjunction
(C)
102,1
Heat(W)
88,4
Code
fins
%ERROR
Bil.Boundary
(C)
Junctiontemperature
algorithmResults
Yovanovich
(C)
rthetaresults
Length(mm)
Full
AluminumExtrusion
Surface
Profiles
HeatTransfer FerrazShawmut/
Mersen
61215
20,7
104
79,6
78,4
33,4%
Width(mm)
41,4
41,4
10
109
88,8
88,8
24,4%
Height(mm)
32,77
62,1
15
115
94,3
95,7
20,1%
82,8
20
119
98,4
101,0
17,2%
Code
66122
21,6
115
93,7
97,8
17,5%
Width(mm)
43,18
43,2
16
131
105,4
115,8
10,2%
Height(mm)
26,42
64,8
24
144
112,5
128,6
7,1%
86,4
32
156
117,6
138,7
6,4%
Code
66454
23,4
10
127
99,3
101,4
24,4%
Width(mm)
46,86
46,9
15
116
95,4
101,9
13,0%
Height(mm)
31,85
70,3
20
118
95,2
105,2
9,8%
93,7
25
121
95,9
108,8
8,1%
fins
fins
fins
Code
66191
25,4
10
131
107,8
118,4
6,5%
Width(mm)
50,8
50,8
15
130
102,2
122,0
130,1 27,8%
8,1%
0,0%
Height(mm)
20,32
76,2
20
137
101,6
128,3
137,9 32,9%
7,8%
1,1%
10
102
25
144
102,1
134,5
144,9 36,5%
8,0%
1,2%
Code
66408
30,5
120
70,9
95,4
23,2%
Width(mm)
60,96
61
122
67,7
98,0
22,7%
Height(mm)
17,78
91,4
12
122
67,2
100,0
21,0%
16
122
15
121
67,4
101,4
19,0%
fins
fins
Code
66419
30,5
25
134
102,7
100,5
30,9%
Width(mm)
61
61
37,5
115
98,8
99,9
14,5%
Height(mm)
71,78
91,5
50
112
99,0
102,8
7,3%
fins
122
62,5
113
100,1
106,4
110,4 15,0%
7,4%
2,6%
66280
36,6
25
140
116,7
114,5
21,1%
Width(mm)
73,1
73,1
37,5
129
111,6
112,4
13,5%
Height(mm)
34,04
110
50
129
111,3
114,6
10,5%
10
146
62,5
132
112,2
117,7
9,9%
Code
fins
43
Bil.boundary
Bil.wall
66195
38
15
121
101,9
110,2
6,4%
Width(mm)
75,95
76
22,5
122
96,5
112,3
3,4%
Height(mm)
16,76
114
30
127
95,8
117,2
2,3%
15
152
37,5
130
96,1
122,2
130,5 34,1%
0,1%
fins
Code
Yovanovich
Code
Bil.Wall
(C)
Bil.Boundary
(C)
%ERROR
Yovanovich
(C)
Junctiontemperature
algorithmResults
Tjunction
(C)
rthetaresults
Heat(W)
AluminumExtrusion
Profiles
FerrazShawmut/
Mersen
Length(mm)
Full
Surface
Heat
Transfer
8,2%
66102
38,1
30
130
108,0
103,8
24,8%
Width(mm)
76,2
76,2
45
115
103,7
102,0
13,4%
Height(mm)
57,15
114
60
113
103,9
104,0
8,3%
152
75
115
105,1
106,8
109,8 11,5%
9,4%
6,0%
Code
66167
38,2
10
104
58,3
93,8
7,0%
Width(mm)
76,48
76,5
15
107
56,5
100,0
104,2 65,5%
8,8%
3,4%
Height(mm)
25,4
115
20
107
56,4
104,2
107,8 65,7%
3,8%
1,0%
25
153
25
106
56,8
107,1
110,3 64,9%
0,9%
5,1%
fins
fins
Code
62350
46,9
10
107
86,7
83,3
30,7%
Width(mm)
93,8
93,8
15
91,9
81,4
78,2
22,0%
Height(mm)
17,75
141
20
86,9
80,3
77,1
17,0%
188
25
85,1
80,1
77,0
77,2
9,1% 14,7%
14,4%
Code
64750
48,1
40
139
113,2
107,1
28,3%
Width(mm)
96,27
96,3
60
118
108,5
103,7
15,2%
Height(mm)
50,8
144
80
114
108,7
104,8
106,2
5,8% 10,5%
8,8%
193
100
113
109,9
106,8
108,5
3,4%
7,1%
5,0%
Code
62705
50,8
40
135
118,3
116,5
14,9%
Width(mm)
101,6
102
60
124
112,9
114,7
6,9%
Height(mm)
33,27
152
80
126
112,6
117,2
121,2 14,1%
9,3%
5,2%
fins
fins
fins
13
203
100
130
113,6
120,7
125,3 16,2%
9,2%
4,5%
Code
66430
55,5
15
124
65,7
113,9
7,3%
Width(mm)
110,9
111
22,5
113
62,9
110,6
112,4 60,5%
3,1%
1,0%
Height(mm)
15,75
166
30
105
62,6
109,4
110,6 56,7%
5,6%
7,2%
fins
36
222
37,5
102
62,8
108,8
109,8 54,5%
9,2%
10,5%
Code
60560
55,6
40
115
109,6
106,8
108,4
6,2%
9,5%
7,6%
Width(mm)
111,1
111
60
108
104,7
104,5
106,7
4,7%
4,9%
2,2%
Height(mm)
32,77
167
80
110
104,6
106,4
109,1
6,3%
4,0%
0,6%
14
222
100
112
105,5
109,2
112,3
7,7%
3,2%
0,7%
Code
61075
57,2
40
122
111,4
107,8
13,2%
Width(mm)
114,3
114
60
109
106,2
105,4
107,6
3,5%
4,6%
1,8%
Height(mm)
35,56
171
80
108
106,1
107,3
110,2
2,5%
0,9%
2,8%
12
229
100
110
107,1
110,2
113,6
3,5%
0,4%
4,6%
Code
66142
58,3
10
106
63,8
89,2
17,4%
Width(mm)
116,6
117
15
96,8
60,6
88,7
9,0%
Height(mm)
13,65
175
20
91,3
60,0
88,6
90,2 51,1%
4,4%
1,9%
27
233
25
87,5
60,0
88,7
89,9 47,8%
2,1%
4,2%
fins
fins
fins
44
63,5
50
118
106,2
101,5
127
127
75
105
102,0
99,1
100,4
4,4%
8,2%
6,5%
Height(mm)
46,02
191
100
105
102,2
100,7
102,4
3,9%
6,0%
3,7%
12
254
125
108
103,5
103,2
105,2
5,1%
5,5%
2,9%
Code
66344
65,8
40
111
94,4
100,5
8,5%
Width(mm)
131,6
132
60
112
90,3
102,9
5,6%
Height(mm)
34,29
197
80
118
90,2
107,5
6,0%
17
263
100
123
91,1
112,1
5,6%
fins
fins
Code
17,2%
Yovanovich
Bil.Wall
(C)
Bil.wall
66414
Width(mm)
Code
Bil.boundary
Bil.Boundary
(C)
%ERROR
Yovanovich
(C)
Junctiontemperature
algorithmResults
Tjunction
(C)
rthetaresults
Heat(W)
AluminumExtrusion
Profiles
FerrazShawmut/
Mersen
Length(mm)
Full
Surface
Heat
Transfer
61790
73
30
124
108,4
103,4
Width(mm)
146
146
45
105
101,1
96,6
96,9
5,0% 11,1%
Height(mm)
17,45
219
60
100
99,7
95,4
95,8
1,0%
7,1%
6,6%
12
292
75
99,2
99,7
95,6
96,0
0,7%
5,2%
4,6%
Code
60230
74,8
50
118
105,6
100,2
19,5%
Width(mm)
149,5
150
75
103
100,7
96,0
96,7
3,7% 10,0%
9,1%
Height(mm)
32,26
224
100
101
100,5
96,5
97,3
1,0%
6,6%
5,5%
14
299
125
102
101,4
98,0
98,9
0,6%
5,5%
4,2%
Code
66451
77,2
100
109
94,4
110,5
116,4 18,0%
2,5%
10,0%
Width(mm)
154,4
154
150
117
92,7
120,1
127,7 27,7%
3,9%
12,6%
Height(mm)
66,56
232
200
126
94,6
129,6
137,7 33,0%
3,4%
11,7%
22
309
250
135
97,3
137,9
145,9 35,7%
3,2%
10,8%
Code
60140
77,2
75
132
117,6
111,1
19,5%
Width(mm)
154,4
154
113
115
112,9
107,7
108,8
2,7%
8,9%
7,5%
Height(mm)
44,45
232
150
112
113,5
109,1
110,5
1,5%
3,7%
2,1%
fins
fins
fins
fins
21,6%
10,6%
13
309
188
114
115,3
111,7
113,3
2,0%
2,2%
0,3%
Code
61070
82,6
80
123
117,1
111,5
112,9
6,7% 12,7%
11,3%
Width(mm)
165,1
165
120
111
112,4
108,8
110,5
2,3%
2,1%
0,0%
Height(mm)
40,64
248
160
109
113,0
111,0
113,1
4,6%
2,1%
4,7%
fins
15
330
200
111
114,9
114,2
116,7
4,3%
3,4%
6,5%
Code
62285
85,7
90
125
120,5
116,6
118,7
4,8%
8,9%
6,6%
Width(mm)
171,5
171
135
113
115,6
115,1
118,0
2,9%
2,3%
5,8%
Height(mm)
41,4
257
180
113
116,3
118,5
122,1
4,3%
6,9%
11,3%
16
343
225
116
118,5
122,9
127,2
3,3%
8,5%
13,5%
Code
66449
88,3
150
122
106,5
123,0
130,0 16,7%
1,3%
8,9%
Width(mm)
176,5
177
225
132
105,3
134,0
142,9 26,5%
1,6%
10,3%
Height(mm)
66,39
265
300
144
108,2
145,3
154,8 31,3%
1,2%
9,6%
25
353
375
153
112,1
155,3
164,8 33,4%
1,7%
9,3%
Code
66279
88,9
80
112
105,3
106,0
108,5
7,9%
7,0%
3,9%
Width(mm)
177,8
178
120
110
101,1
106,5
110,0 11,4%
4,7%
0,3%
Height(mm)
33,32
267
160
114
101,7
110,5
114,8 15,1%
4,6%
0,5%
23
356
200
120
103,4
115,2
120,0 18,1%
5,0%
0,4%
fins
fins
fins
45
Bil.boundary
66288
102
40
121
71,7
118,6
123,1 54,0%
2,4%
2,5%
Width(mm)
203,2
203
60
111
68,8
116,4
119,0 52,0%
7,0%
10,2%
Height(mm)
20,32
305
80
105
68,8
116,0
17,2%
40
406
100
101
69,6
116,1
23,3%
Code
66226
105
100
111
104,2
102,9
104,7
Width(mm)
209,6
210
150
109
100,4
102,6
105,1 10,3%
7,5%
4,3%
Height(mm)
33,58
314
200
113
101,4
106,3
109,3 14,0%
8,1%
4,4%
26
419
250
118
103,6
110,6
114,1 16,6%
8,6%
4,6%
Code
60815
106
100
135
113,9
106,5
26,4%
Width(mm)
212,1
212
150
114
109,6
103,0
103,5
5,2% 13,1%
12,6%
Height(mm)
50,8
318
200
109
110,6
104,4
104,9
1,6%
6,2%
5,6%
12
424
250
109
112,8
106,9
107,5
5,4%
2,2%
1,5%
Code
66179
114
125
117
94,1
124,7
133,1 26,0%
9,3%
18,9%
Width(mm)
228,6
229
188
123
92,2
134,1
20,8%
Height(mm)
39,88
343
250
127
94,3
141,8
22,4%
36
457
313
130
97,3
148,1
24,0%
Code
66180
114
150
122
98,0
124,8
132,5 25,8%
3,5%
11,9%
Width(mm)
228,6
229
225
129
96,5
134,0
141,2 33,0%
4,7%
11,9%
Height(mm)
41,15
343
300
135
99,1
142,0
148,3 34,1%
6,8%
12,7%
40
457
375
139
102,7
148,8
154,2 33,4%
8,9%
13,8%
Code
66434
122
150
104
85,0
111,1
118,4 25,5%
9,8%
19,7%
Width(mm)
243,9
244
225
111
84,7
122,5
23,4%
Height(mm)
49,02
366
300
118
87,7
132,5
24,9%
26,7%
fins
fins
fins
fins
fins
fins
8,5% 10,2%
Bil.wall
Yovanovich
Code
Bil.Wall
(C)
Bil.Boundary
(C)
%ERROR
Yovanovich
(C)
Junctiontemperature
algorithmResults
Tjunction
(C)
rthetaresults
Heat(W)
AluminumExtrusion
Profiles
FerrazShawmut/
Mersen
Length(mm)
Full
Surface
Heat
Transfer
7,9%
44
488
375
123
91,4
140,8
Code
66221
124
100
115
110,9
111,0
113,5
5,0%
4,8%
1,8%
Width(mm)
247,7
248
150
112
105,4
109,6
112,8
8,5%
3,3%
0,5%
Height(mm)
20,07
371
200
115
105,7
113,0
116,7 10,9%
2,4%
2,0%
121,3 10,8%
fins
32
495
250
117
107,6
117,2
0,2%
4,9%
Code
62725
124
120
92,6
92,8
88,4
89,1
0,3%
6,7%
5,5%
Width(mm)
247,7
248
180
87,1
91,0
88,6
89,6
6,9%
2,6%
4,4%
Height(mm)
57,91
371
240
89,4
93,1
92,0
93,3
6,2%
4,4%
6,6%
20
495
300
93,1
96,0
96,2
97,7
4,6%
4,9%
7,3%
Code
60520
125
150
119
114,9
122,0
126,9
5,0%
3,0%
8,5%
Width(mm)
250,8
251
225
120
111,4
126,0
132,1
9,9%
6,3%
13,1%
Height(mm)
34,29
376
300
127
113,4
133,1
139,9 13,7%
6,8%
13,8%
30
502
375
132
116,7
140,4
147,4 15,0%
8,2%
15,1%
Code
66428
131
200
120
115,4
108,8
109,7
5,0% 12,3%
11,3%
Width(mm)
262,3
262
300
111
114,1
109,3
110,4
4,2%
1,8%
0,4%
Height(mm)
60,93
393
400
113
117,8
114,2
115,6
5,3%
0,9%
2,6%
21
525
500
118
122,7
120,1
121,8
5,1%
2,1%
4,1%
fins
fins
fins
46
Yovanovich
Bil.boundary
Bil.wall
Code
66143
133
150
114
107,0
111,3
114,8
8,2%
3,1%
1,0%
Width(mm)
265,9
266
225
115
104,2
114,5
118,9 12,3%
0,1%
5,1%
Height(mm)
41,28
399
300
120
106,2
120,7
125,7 15,6%
0,5%
6,0%
28
532
375
127
109,6
127,2
132,5 17,6%
0,6%
6,1%
Code
66395
136
150
133
132,0
126,0
127,2
0,7%
6,6%
5,4%
Width(mm)
272,4
272
225
122
126,0
121,4
122,7
4,8%
0,2%
1,2%
Height(mm)
26,67
409
300
123
127,0
123,4
124,8
4,9%
1,0%
2,5%
23
545
375
126
129,8
127,1
128,6
4,2%
1,3%
3,0%
Code
65340
140
200
110
112,5
109,5
111,4
3,7%
0,0%
2,4%
Width(mm)
279,4
279
300
108
111,3
112,2
114,9
3,7%
4,8%
8,2%
Height(mm)
58,42
419
400
115
115,0
118,8
122,2
0,2%
4,7%
8,7%
23
559
500
122
119,8
126,1
130,2
2,5%
4,3%
8,7%
14,9%
fins
fins
fins
Code
Bil.Wall
(C)
Bil.Boundary
(C)
%ERROR
Yovanovich
(C)
Junctiontemperature
algorithmResults
Tjunction
(C)
rthetaresults
Heat(W)
AluminumExtrusion
Profiles
FerrazShawmut/
Mersen
Length(mm)
Full
Surface
Heat
Transfer
66427
155
250
126
118,7
111,1
111,7
7,6% 15,6%
Width(mm)
310
310
375
115
118,6
112,0
112,7
4,5%
3,3%
2,5%
Height(mm)
74,3
465
500
117
123,5
117,6
118,4
8,1%
1,3%
2,2%
18
620
625
122
129,6
124,2
125,2
8,7%
2,9%
3,9%
Code
61155
156
150
123
116,6
124,9
130,3
6,5%
2,4%
8,2%
Width(mm)
311,2
311
225
123
112,3
128,1
134,5 11,8%
5,1%
12,0%
Height(mm)
25,4
467
300
126
113,9
134,8
141,8 12,3%
9,5%
16,8%
32
622
375
128
117,0
141,8
21,5%
Code
66459
170
200
112
106,3
115,0
119,6
6,7%
3,9%
9,5%
Width(mm)
339,9
340
300
116
104,7
120,7
126,3 12,6%
6,1%
12,7%
Height(mm)
33,32
510
400
121
108,0
129,0
135,0 14,4%
8,6%
15,2%
40
680
500
126
112,6
137,1
18,0%
fins
fins
fins
Iftheerrorisbelow15%,theerrorishighlightedwithboldtypography.Contrary,inerrorsexceeding
15%thenumberiscoloredred.InallprofilesisconsideredaRjcresistanceof0.05C/W
47
Appendix2.AIRPROPIETIESANDEQUATIONS
Tableofairproprieties
Temperature
Density
Volumetric
thermal
expansivity
Heat
Capacity
Conductivity
Dynamic
Viscosity
Kinematic
viscosity
Prandtl
Cp
Pr
10
10
10
[C]
[F]
[K]
32
273
1,293
3,664
1003,9
0,02417
17,17
13,28
0,713
41
278
1,269
3,598
1004,3
0,02445
17,35
13,67
0,713
10
50
283
1,242
3,533
1004,6
0,02480
17,58
14,16
0,712
15
59
288
1,222
3,470
1004,9
0,02512
17,79
14,56
0,712
20
68
293
1,202
3,412
1005,2
0,02544
18,00
14,98
0,711
25
77
298
1,183
3,354
1005,4
0,02577
18,22
15,40
0,711
30
86
303
1,164
3,298
1005,7
0,02614
18,46
15,86
0,710
35
95
308
1,147
3,244
1006,0
0,02650
18,70
16,30
0,710
40
104 313
1,129
3,193
1006,3
0,02684
18,92
16,76
0,709
45
113 318
1,111
3,142
1006,6
0,02726
19,19
17,27
0,709
50
122 323
1,093
3,094
1006,9
0,02761
19,42
17,77
0,708
55
131 328
1,079
3,048
1007,3
0,02801
19,68
18,24
0,708
60
140 333
1,061
3,003
1007,7
0,02837
19,91
18,77
0,707
65
149 338
1,047
2,957
1008,0
0,02876
20,16
19,26
0,707
70
158 343
1,030
2,914
1008,4
0,02912
20,39
19,80
0,706
75
167 348
1,013
2,875
1008,8
0,02945
20,60
20,34
0,706
80
176 353
1,001
2,834
1009,3
0,02979
20,82
20,80
0,705
85
185 358
0,986
2,795
1009,8
0,03012
21,02
21,32
0,705
90
194 363
0,972
2,755
1010,3
0,03045
21,23
21,84
0,704
95
203 368
0,959
2,718
1010,7
0,03073
21,41
22,33
0,704
0,947
2,683
1011,2
0,03101
21,58
22,79
0,704
48
Proprietiesequations
Fittedcurvesforairproprieties.Validfrom0to100C.
Equations.TemperatureTinCelsiusdegrees.
9.8618 10
3.1601 10
4.3945 10
1.2878 10
1.2884
3.659 10
4.3574 10
1.3179 10
1.5635 10
9.4276 10
2.8071 10
9.2169 10
1003.9
2.7725 10
4.0404 10
2.4180 10
2.6013 10
1.8732 10
17.173 10
4.63522 10
6.09812 10
7.13087 10
7.86027 10
0,99992
0,99978
0.99995
0,99996
0,99996
0,99992
3.4864 10
5.4658 10
9.6597 10
8.4954 10
13.278 10
0,99966
5.5634 10
3.679 10
R2
Units
R2isthesquarerootofthecorrelationcoefficient.
49
Appendix3.MultipleheatsourcealgorithmforMatlab.
Naturalconvectionalgorithm
function [Tj]=dissipateur
(Ld,wd,xd,yd,P,Rjc,Tamb,L,w,tp,Slibre,k,H,tab,taf,sb,na,Sem)
%Function.ThisfunctionacceptsLd,wd,xd,yd,P,Rjc,Tamb,L,w,tp,Slibre,k,H,tab,taf,sb,na,Sem
variablesandreturnsthejunctiontemperaturevector[Tj].Assample,allpreviousaredefined,takingas
modelthe64750profile.3Heatsourcesarealsodefined.
%Deviceparameters
DeviceLengthvector[m]
%Ld=[0.040.040.04];
%wd=[0.0250.0250.025];
DeviceWidthvector[m]
%xd=[0.050.050.05];
Devicexpositionvector[m]
%yd=[0.0750.150.225];
Devicexpositionvector[m]
%P=[606060];
Deviceheatrategeneration[W]
%Rjc=[0.050.050.05];
ResistanceJunctioncase[C/W]
%HeatSinkparameters
%Sem=0.77;
Surfaceemmisivity
%L=0.3;
HeatsinkLength[m]
%w=0.09627;
Heatsinkwidth[m]
%tp=0.00508;
Platethickness[m]
%Slibre=0
Partodtheplatewidthwithoutfins[m]
%k=210;
ConductivityofthematerialoftheHeatSink[W/mK]
%H=0.046;
Finheight[m]
%tab=3.466E3;
Finbasethickness[m]
%taf=2.124E3;
Fintipthickness[m]
Interfinspaceatbase[m]
%sb=8.135E3;
%na=9;
Numberoffins
%AmbientParameter
%Tamb=30;
Ambienttemperature[C]
Ab=w*L;
%Sinktotalsurface[m2]
%Airproprieties
Tm=@(Ts) (Ts+Tamb)/2;
%Filmtemperature[C]
Betaa=@(T) 1/(T+273);
%Volumetricthermalexpansivity[1/K]
Prandtl=@(T) 4.63522E-9*T^3-6.09812E-7*T^2-7.86027E-5*T+7.13087E-1; %Prandtl
numberfunction
Kinvis=@(T) (-3.4864E-8*T^4+5.4658E-6*T^3-9.6597E-5*T^2+8.4954E2*T+1.3278E1)*1E-6; %Kinematicviscosity[m2/s]
Dynvis=@ (T) (-1.8732E-6*T^3+2.6013E-4*T^2+3.679E-2*T+17.173)*1E-6; %Dynamic
viscosity[Pas]
ka=@(T) -2.7725E-9*T^3+4.0404E-7*T^2+5.5634E-5*T+2.418E-2; %Airthermal
conductivity[W/mK]alpha=@(T)Kinvis(T)/Prandtl(T); %Thermaldiffusivity[m2/s]
Dens=@ (T) 9.8618E-6*T^2-4.3945E-3*T+1.2884; %Thermaldensity[Kg/m3]
Cp=@ (T) -4.3574E-11*T^6+1.3179E-8*T^5-1.5635E-6*T^4+9.4276E-5*T^3-2.8071E3*T^2+9.2169E-2*T+1.0039E3; %AirHeatCapacity[Kg/m3]
50
%Totalpowergenerated[W]
Nd=length(Ld);
Ptot=0;
for b=1:Nd
Ptot=P(b)+Ptot;
end
Hc=H+taf/2;
%Finheightcorrection[m]
Ap=(Slibre+(na-1)*sb)*L;
%Primarysurface[m2]
Af=Hc*2*na*L;
%Finsurface[m2]
kappa=atan((tab-taf)/(2*H)); %Finangle[rad]
sm=(2*sb+tab-taf)/2;
%AverageInterfinspace[m]
g=9.81;
%Gravity[m/s2]
sigma=5.6704E-8;
%StefanBoltzmannconstant
%Finefficiencyforatrapezoidalprofile
K=@(h) sqrt(h/(k*sin(kappa)));
mua=@(h) 2*K(h)*sqrt(taf*(1-tan(kappa))/(2*tan(kappa)));
mub=@(h) 2*K(h)*sqrt(H+taf*(1-tan(kappa))/(2*tan(kappa)));
eff=@(h) (mub(h)/(2*H*(K(h))^2))*(besselk(1,mua(h))*besseli(1,mub(h))besseli(1,mua(h))*besselk(1,mub(h)))/(besseli(0,mub(h))*besselk(1,mua(h))+b
esseli(1,mua(h))*besselk(0,mub(h)));
%CONVECTIONCOEFFICIENT.CHOICEONEOFTHISTHREEPROPOSEDCONVECTIONMODELS.
%CONVECTIONCOEFFICIENTBYYOVANOVICH
S=(na*(2*(((tab+taf)/2)^2+H^2)^0.5+taf)+(w-na*tab))*L;
RaS=@(Ts) (g*Betaa(Tm(Ts))*Prandtl(Tm(Ts))*(TsTamb)*S^1.5)/(Kinvis(Tm(Ts))^2);
%RaSRayleichnumber
f=@ (Ts) 0.67/(1+(0.5/Prandtl(Tm(Ts)))^(9/16))^(4/9);
%UniversalPrandtl
function
Nudl=(3.192+1.868*(H/L)^0.76)/(1+1.189*(H/L))^0.5;
%Diffusivelimit
Lambda=na*H+tp+w;
GS=2^(1/8)*((L*(H*na+tp+w)^2)/((tab/2+taf/2)*H*na+tp*w+L*(H*na+tp+w))^1.5)^
0.25;
%Body
function
NuS=@(Ts) Nudl+f(Ts)*GS*(RaS(Ts))^0.25;
%Nusseltnumber
hc=@ (Ts) ka(Tm(Ts))*NuS(Tm(Ts))/(S^0.5);
%Convection
coefficient
%CONVECTIONCOEFFICIENTBYBILITZKYATTs
r=2*H*sm/(2*H+sm);
a=sm/H;
B=1.25*(1+sm/(2*H));
Lambda1=1-0.483*exp(-0.17/a);
Lambda2=1-exp(-0.83*a);
Lambda3=9.14*(a.^0.5)*exp(-B)-0.61;
Psi=24*Lambda1/(((1+a/2)*(1+Lambda2*Lambda3)).^3);
Rar=@(Ts) ((Dens(Ts)^2)*g*Betaa(Tm(Ts))*Cp(Ts)*(TsTamb)*r^3)/(Dynvis(Ts)*ka(Ts));
El=@(Ts) Rar(Ts)*r/L;
%RaSRayleichnumber
%Elenbaasnumber
51
Nur=@(Ts) El(Ts)*(1-exp(-Psi*((0.5/El(Ts)).^0.75)))/Psi;
hc= @(Ts) Nur(Ts)*ka(Ts)/r;
coefficient
%Nusseltnumber
%Convection
%CONVECTIONCOEFFICIENTBYBILITZKYATTm
r=2*H*sm/(2*H+sm);
a=sm/H;
B=1.25*(1+sm/(2*H));
Lambda1=1-0.483*exp(-0.17/a);
Lambda2=1-exp(-0.83*a);
Lambda3=9.14*(a.^0.5)*exp(-B)-0.61;
Psi=24*Lambda1/(((1+a/2)*(1+Lambda2*Lambda3)).^3);
Rar=@(Ts) ((Dens(Tm(Ts)^2)*g*Betaa(Tm(Ts))*Cp(Tm(Ts)*(TsTamb)*r^3)/(Dynvis(Tm(Ts)*ka(Tm(Ts));
El=@(Ts) Rar(Ts)*r/L;
Nur=@(Ts) El(Ts)*(1-exp(-Psi*((0.5/El(Ts)).^0.75)))/Psi;
hc= @(Ts) Nur(Ts)*ka(Ts)/r;
coefficient
%RaSRayleichnumber
%Elenbaasnumber
%Nusseltnumber
%Convection
%RADIATIONCALCULATION
Hr=H/sm;
Lr=L/sm;
Fs=1-2*Hr*((1+Lr.^2).^0.5-1)/(2*Hr*Lr+(1+Lr.^2).^0.5-1);
Qch=@(Ts) sigma*(sm+2*H)*L*((Ts+273.15).^4-(Tamb+273.15).^4)/(((1Sem)/Sem)+1/Fs);
Qtr=@(Ts)
na*Qch(Ts)+(na*tp*(L+2*H)+2*H*L+2*tp*(L+w))*sigma*Sem*((Ts+273).^4(Tamb+273.15).^4);
Ahs=Ap+Af+tp*2*(w+L)+na*(tab+taf)*(2*H+L)/2;
hr=@(Ts) Qtr(Ts)/(Ahs*(Ts-Tamb));
%SPREADINGCALCULATION
Phi=@(zeta,h)
((zeta*sinh(zeta*tp))+(h*cos(zeta*tp)/k))/((zeta*cosh(zeta*tp))+(h*sinh(zet
a*tp)/k));
%ITERATIONPROCESS
%Tsevaluation
Pnouvell=0.1;
j=1;
Error=100;
Ts=Tamb+90;
iter=1;
while Error>0.001
h=hr(Ts)+hc(Ts);
Ts1=(Ptot/(h*(Ap+eff(h)*Af)))+Tamb;
Error=abs(Ts1-Ts);
Ts=Ts1;
iter=iter+1;
end
%EvaluationoftheTemperaturedropvector
52
%ITERATIONPROCESS
hm=(Ap+eff(h)*Af)*h/Ab;
for j=1:Nd
theta1(j)=0;
for i=1:Nd
A0(i)= P(i)*((tp/k)+(1/hm))/(w*L);
sum1=0;
sum2=0;
sum3=0;
for m=1:100
lambda=(m*pi)/w;
Am=2*P(i)*(sin(((2*xd(i)+wd(i))*lambda/2))-sin((2*xd(i)wd(i))*(lambda/2)))/(w*L*wd(i)*k*(lambda.^2)*Phi(lambda,hm));
sum1=(Am*cos(lambda*xd(j))*sin(lambda*wd(j)/2)/(lambda*wd(j)))+sum1;
end
for n=1:100
delta=(n*pi)/L;
An=2*P(i)*(sin((2*yd(i)+Ld(i))*delta/2)-sin((2*yd(i)Ld(i))*delta/2))/(w*L*Ld(i)*k*(delta.^2)*Phi(delta,hm));
sum2=(An*cos(delta*yd(j))*sin(delta*Ld(j)/2))/(delta*Ld(j))+sum
2;
end
for m=1:100
for n=1:100
lambda=(m*pi)/w;
delta=(n*pi)/L;
beta=sqrt((lambda.^2)+(delta.^2));
Amn=(16*P(i)*cos(lambda*xd(i))*sin(lambda*wd(i)/2)*cos(de
lta*yd(i))*sin(delta*Ld(i)/2))/(w*L*wd(i)*Ld(i)*k*beta*la
mbda*delta*Phi(beta,hm));
sum3=(Amn*cos(delta*yd(j))*sin(delta*Ld(j)/2)*cos(lambda*
xd(j))*sin(lambda*wd(j)/2))/(lambda*wd(j)*delta*Ld(j))+su
m3;
end
end
theta(i,j)=A0(i)+2*sum1+2*sum2+4*sum3;
end
for i=1:Nd
theta1(j)=theta(i,j)+theta1(j);
end
end
%Junctiontemperaturevectorevaluation
for j=1:Nd
Tc(j)=theta1(j)+Tamb;
Tj(j)=Tc(j)+P(j)*Rjc(j);
end
53